Added iogalle

Author: cooper-mj

.DS_Store

@@ -0,0 +1,80 @@
+WAGE PREDICTOR
+==============
+VIDEO LINK: https://photos.app.goo.gl/VJ6oWVugFbgz1XhAA
+
+--------
+Overview
+--------
+My project fits a linear model to an online dataset containing information
+about people's wages, height, sex, race, education level, and age. The predict_wages
+module processes the data and fits the model, and the wages_console allows users
+to interact with the model and enter their own information to get a prediction.
+This project also calculates how the wage predictions change if a person's sex
+or race were different.
+
+-----------------
+Technical Details
+-----------------
+The predict_wages module processes the data by reading in a csv file, and then cleans
+the data by separating the wage information from the feature information, and replaces
+the categorical race and gender information with one-hot encoding. Finally, it
+uses least squares to fit a model to the data. The wages_console prompts the user
+for input, and uses this input to create a new user feature vector. Using the model
+created by the predict_wages module, we make a wage prediction for the new user.
+
+-------------
+Prerequisites
+-------------
+numpy, pandas, random
+
+----------------
+Running the Code
+----------------
+Run the wages_console.py script with "python3 wages_console.py". This program will
+fit the model to the data, then prompt the user to enter input, and finally report
+its predictions.
+
+------------
+Example Test
+------------
+    $ python3 wages_console.py
+
+        Welcome to the Wage Predictor!
+        Enter some information about yourself, and we'll predict your wage.
+        Then see how your wage would change if your gender or race were different.
+
+
+    Creating model...
+    Model fitted.
+
+    ----------------------------------
+    Please enter your height in inches. 61
+    Please enter your sex. female
+    Please enter your race. Hispanic
+    Please enter the number of years you attended school. 14
+    Please enter your age in years. 20
+
+    We predict that your wage will be $13789.06.
+
+    If you were male, your predicted wage would be $17552.54 higher.
+    If you were white, your predicted wage would be $22145.33 higher.
+
+    Again? (Y/N) N
+    Goodbye!
+
+----------
+Known Bugs
+----------
+Unfortunately I had some trouble finding datasets that fit the needs of this
+project, so this is the only dataset containing wage, race, and gender information
+that I could find. The dataset didn't have much documentation, so I have no idea
+what the wage values mean (e.g. annual income?). Also, the linear model may not be
+the best fit, because sometimes the model predicts huge negative values, which doesn't
+seem to make too much sense...
+Further testing required!
+
+-------
+Authors
+-------
+I completed this assignment independently, but I used some of the code from
+Assignments 1 and 2 as a template.

iogalle/README.md

@@ -0,0 +1,137 @@
+#!/usr/bin/env python3
+"""
+Name: Isabel Gallegos
+Creates a linear model of the relationship between height, sex, race, education
+level, and age and expected wage.
+"""
+import numpy as np
+import pandas
+
+INPUT_FILENAME = 'wages.csv'
+
+def load_data(filename):
+    """
+    Loads the wage data from a csv file.
+
+    Arguments:
+    filename -- the file that contains the wage data
+
+    Returns:
+    wage_data -- numpy array with the following fields
+        "earn"
+        "height"
+        "sex"
+        "race"
+        "ed"
+        "age"
+    """
+    wage_data = pandas.read_csv(filename, delimiter = ',').to_numpy()
+    return wage_data
+
+
+def clean_data(wage_data):
+    """
+    Cleans the wage data into two numpy arrays.
+
+    Arguments:
+    wage_data -- a numpy array of wages and features (height, sex, race, ed, age)
+
+    Returns:
+    X -- a numpy array of length 1379 x 5 where each row has the form
+         [height, sex, race, ed, age]
+    y -- a 1379-length numpy array, where y[i] is the wage associated with X[i]
+    """
+    # get wages
+    y = wage_data[:,0]
+    y = y.reshape(y.size, 1)
+    y = y.astype(float)
+
+    # get features, convert strings to 0/1 values
+    height = wage_data[:,1]
+    height = height.reshape(height.size, 1)
+
+    sex = wage_data[:,2]
+    male = np.where(sex == "male", 1, 0)
+    male = male.reshape(male.size, 1)
+    female = np.where(sex == "female", 1, 0)
+    female = female.reshape(female.size, 1)
+
+    race = wage_data[:,3]
+    white = np.where(race == "white", 1, 0)
+    white = white.reshape(white.size, 1)
+    black = np.where(race == "black", 1, 0)
+    black = black.reshape(black.size, 1)
+    hispanic = np.where(race == "hispanic", 1, 0)
+    hispanic = hispanic.reshape(hispanic.size, 1)
+    other = np.where(race == "other", 1, 0)
+    other = other.reshape(other.size, 1)
+
+    ed = wage_data[:,4]
+    ed = ed.reshape(ed.size, 1)
+    age = wage_data[:,5]
+    age = age.reshape(age.size, 1)
+
+    X = np.concatenate((height, male, female, white, black, hispanic, other, ed, age), axis=1)
+    X = X.astype(float)
+
+    return X, y
+
+
+def fit_model(X, y):
+    """
+    Processes the wage data by fitting a line to it.
+
+    Arguments:
+    X -- an 1379 x 9 numpy array, where each row is of the form [height, male,
+         female, white, black, hispanic, other, ed, age] representing a single data point
+    y -- a 800-length numpy array, where y[i] is the wage for the individual
+         associated with X[i]
+
+    Returns:
+    weights of the model
+    """
+    return np.linalg.lstsq(X, y, rcond=None)[0]
+
+
+def create_model():
+    """
+    Loads and cleans data, and uses linear regression to model relationship
+    between the features [height, male, female, white, black, hispanic, other, ed, age]
+    and wage.
+
+    Returns:
+    weights -- weights of the model
+    """
+    print("\nCreating model...")
+    # Load data
+    wage_data = load_data(INPUT_FILENAME)
+
+    # Inform user about data
+    if wage_data is None:
+        print("Warning: no data recieved.")
+        return
+
+    # Clean data
+    clean_data(wage_data)
+    try:
+        X, y = clean_data(wage_data)
+    except TypeError:
+        X, y = None, None
+
+    # Inform user about data
+    if X is None and y is None:
+        print("Warning: no data cleaned.")
+        return
+
+    # Process data
+    try:
+        weights = fit_model(X, y)
+    except TypeError:
+        weights = None
+
+    if weights is None:
+        print("Warning: no model fitted.")
+        return
+
+    print("Model fitted.\n")
+    return weights